US12392660B1ActiveUtility
Curve-based radiometric calibration method and system of spaceborne hyperspectral imager
Est. expiryMay 8, 2044(~17.8 yrs left)· nominal 20-yr term from priority
G01J 3/28G01J 3/0297G01J 2003/2826G01J 2003/2879G01J 3/2823G01J 3/2803
53
PatentIndex Score
0
Cited by
9
References
10
Claims
Abstract
The present disclosure provides a curve-based radiometric calibration method and system of a spaceborne hyperspectral imager, which belongs to the field of remote-sensing optical technologies. A new radiometric calibration coefficient curve is introduced to describe the radiometric properties of sensor response, providing the radiometric calibration coefficients of all bands of the hyperspectral camera with linear variable filter (LVF) within an imaging spectral range to match the implementation of the programmable band selection imaging technology, and thereby efficiently and simply implementing single-band imaging, integral imaging of adjacent band, imaging of randomly-selected band combination and within-spectral-range cyclic imaging. In the present disclosure, the radiometric calibration coefficient is used to cover the entire imaging spectral range of the hyperspectral camera and match the implementation of the programmable band selection imaging technology, and realizing on-orbit absolute radiometric calibration with simple flow and strong universality.
Claims
exact text as granted — not AI-modifiedWhat is claimed is:
1. A curve-based radiometric calibration method of a spaceborne hyperspectral imager, comprising:
before a to-be-determined hyperspectral imager is launched, matching a linear variable filter with a Complementary Metal Oxide Semiconductor (CMOS) detector array to acquire a matching result of a CMOS detector row number;
in a laboratory dark background environment, performing no-reference radiation source imaging and calibration light source imaging of the to-be-determined hyperspectral imager, and completing additive noise correction and non-uniformity correction of the CMOS detector;
based on the calibration source light imaging and measured spectral radiance distribution, measuring a radiometric calibration coefficient reference value of several conventional spectral channels;
based on the radiometric calibration coefficient reference value, establishing a prelaunch radiometric calibration coefficient curve equation and determining a radiometric calibration curve accuracy evaluation result;
after the to-be-determined hyperspectral imager is launched, establishing a radiometric reference transmission link;
with constraints of spatial, spectral and temporal resolutions, completing system-level preprocessing of the to-be-determined hyperspectral imager and measuring a radiometric calibration coefficient reference value of the to-be-determined hyperspectral imager on orbit;
based on the radiometric calibration curve accuracy evaluation result, performing fitting on data points of the radiometric calibration coefficient reference value by using a fitting algorithm and establishing a postlaunch radiometric calibration coefficient curve equation;
based on the postlaunch radiometric calibration coefficient curve equation, performing periodic calculation on the radiometric calibration coefficient reference value of the to-be-determined hyperspectral imager on orbit to obtain an updated on-orbit attenuated radiometric calibration curve, and obtaining an imaging band radiometric calibration coefficient based on the updated on-orbit attenuated radiometric calibration curve.
2. The curve-based radiometric calibration method of the spaceborne hyperspectral imager according to claim 1 , wherein before the to-be-determined hyperspectral imager is launched, matching the linear variable filter with the CMOS detector array to acquire the matching result of the CMOS detector row number comprises the following steps:
at step 1.1, with a monochromator as a wavelength-continuously-adjustable light source irradiating the to-be-determined hyperspectral imager through a collimator, the monochromator, with a minimum step distance as a start of wavelength change, repetitively outputs actual stepping corresponding to each standard wavelength by starting from a start wavelength of the to-be-determined hyperspectral imager, and collects calibration images of different wavelengths output by the to-be-determined hyperspectral imager until an ending wavelength of the to-be-determined hyperspectral imager is reached;
at step 1.2, a mean gray value of an image with a central wavelength of λ c is calculated row-by-row and a row with a maximum mean gray value is used as a corresponding output row j of the central wavelength λ c ;
at step 1.3, the step 1.2 is repeated within an imaging spectral range to obtain corresponding output rows of different central wavelengths, and each central wavelength and a row number corresponding to the maximum on the corresponding calibration image are determined;
at step 1.4, based on least square method, a linear relational expression λ c (j)=g×j+w 0 between CMOS detector row number and central wavelength is established, wherein g is a wavelength change gradient of the linear variable filter, and w 0 is a wavelength corresponding to a start mechanical position of the linear variable filter.
3. The curve-based radiometric calibration method of the spaceborne hyperspectral imager according to claim 1 , wherein in the laboratory dark background environment, performing the no-reference radiation source imaging and the calibration the light source imaging of the to-be-determined hyperspectral imager, and completing the additive noise correction and the non-uniformity correction of the CMOS detector comprise the following steps:
at step 2.1, the to-be-determined hyperspectral imager is enabled to work in a dark background imaging mode to collect dark background image data several times;
at step 2.2, the to-be-determined hyperspectral imager is set to default imaging parameters, the to-be-determined hyperspectral imager is irradiated with an integrating sphere to collect valid image data several times, and a spectral radiance distribution of the integrating sphere is measured with a spectroradiometer;
at step 2.3, the spectral radiance distribution of the integrating sphere is continuously and uniformly changed several times and the step 2.2 is repeated;
at step 2.4, based on the collected calibration images and the spectral radiance distribution output by the integrating sphere, a relative radiometric calibration coefficient and a relative response non-linearity of the to-be-determined hyperspectral imager are calculated:
DN ι,k − DN ι,0 =( DN i,j,k −DN i,j,0 )* a i,j +b i,j
wherein DN ι,k is an output signal mean value of the i-th row of valid detector elements under the k-th-level radiance, DN ι,0 is an output signal mean value of the i-th row of valid detector elements at the time of no-reference radiation source, DN i,j,k is an output signal mean value of the picture element of the i-th row and the j-th column under the k-th-level radiance, DN i,j,0 is an output signal of the picture element of the i-th row and the j-th column at the time of no-reference radiation source, a i,j , b i,j are a relative radiometric calibration coefficient and a relative radiometric calibration constant of the j-th detector element of the i-th band.
4. The curve-based radiometric calibration method of the spaceborne hyperspectral imager according to claim 1 , wherein based on the calibration source light imaging and the measured spectral radiance distribution, measuring the radiometric calibration coefficient reference value of several conventional spectral channels comprises the following steps:
at step 3.1, based on the collected calibration images and the spectral radiance distribution output by the integrating sphere, an absolute radiometric calibration coefficient of each spectral channel is calculated, and the calculation of an equivalent radiance of the to-be-determined hyperspectral imager corresponding to the spectral radiance distribution output by the integrating sphere comprises:
L
e
=
∫
λ
min
λ
max
L
λ
R
λ
d
λ
∫
λ
min
λ
max
R
λ
d
λ
wherein L e is an equivalent radiance of a current spectral channel, R λ is a relative spectral response function of the current spectral channel, λ min is a minimum wavelength of the relative spectral response function, λ max is a maximum wavelength of the relative spectral response function, and L λ is a spectral radiance distribution output by the integrating sphere;
at step 3.2, an output row j corresponding to the central wavelength λ c of the current spectral channel is selected on the current calibration image, and a mean gray value L j of all pixels on the row j is calculated;
at step 3.3, the calibration images corresponding to different spectral radiance distributions are selected to repeat the step 3.2 so as to obtain corresponding mean gray values; through relative radiometric correction and additive noise elimination, the formula L λ =G λ · L j is obtained, wherein L λ is a spectral radiance distribution output by the integrating sphere, L j is the mean gray value of all pixels on the j-th row of a camera detector, and G λ is a radiometric calibration coefficient reference value; by least square fitting, the radiometric calibration coefficient of any spectral channel is calculated;
at step 3.4, the calibration images of different spectral channels are selected to repeat the step 3.3 so as to obtain the radiometric calibration coefficients of several conventional spectral channels as the radiometric calibration coefficient reference values.
5. The curve-based radiometric calibration method of the spaceborne hyperspectral imager according to claim 1 , wherein based on the radiometric calibration coefficient reference value, establishing the prelaunch radiometric calibration coefficient curve equation and determining the radiometric calibration curve accuracy evaluation result comprise the following steps:
at step 4.1, based on the linear relational expression λ c (j)=g×j+w 0 between CMOS detector row number and central wavelength and the radiometric calibration coefficient reference value G λ , regression analysis is performed on multiple groups of data points to obtain a curve equation of radiometric calibration coefficient reference value G λ =F(j), wherein G λ is a radiometric calibration coefficient reference value, j is a detector array row number, and F is a radiometric calibration coefficient curve equation;
at step 4.2,
R
2
=
1
-
∑
λ
=
1
n
(
M
λ
-
F
λ
)
2
∑
λ
=
1
n
(
M
λ
-
M
_
)
2
is determined as the radiometric calibration curve accuracy evaluation result, wherein F λ is a fitting value with the band central wavelength being λ, M λ is a measurement reference value with the band central wavelength being λ, n is a total wavelength number, R 2 is a determination coefficient, and M is a measurement reference mean value with the band central wavelength being A;
at step 4.3, based on the radiometric calibration curve accuracy evaluation result, a polynomial function is used as the prelaunch radiometric calibration coefficient curve equation:
G λ =p 1 λ j n +p 2 λ j n-1 + . . . +p n λ j +p n+1
wherein n is a degree of a polynomial, namely, a total wavelength number, and p 1 , p 2 . . . , p n+1 are coefficients of the degrees of the corresponding items of the polynomial.
6. The curve-based radiometric calibration method of the spaceborne hyperspectral imager according to claim 1 , wherein after the to-be-determined hyperspectral imager is launched, establishing the radiometric reference transmission link comprises the following steps:
at step 5.1, integration is performed on reference Top-of-Atmosphere reflectance data ρ TOA (λ) based on spectral integration formula to obtain a mean Top-of-Atmosphere reflectance ρ TOA (λ) corresponding to the central wavelength λ of the to-be-determined hyperspectral imager:
〈
ρ
TOA
(
λ
)
〉
=
∫
λ
∈
B
ρ
TOA
(
λ
)
S
B
(
λ
)
d
λ
∫
λ
∈
B
S
B
(
λ
)
d
λ
wherein ρ TOA (λ) is a mean Top-of-Atmosphere reflectance corresponding to the central wavelength λ of the to-be-determined hyperspectral imager, ρ TOA (λ) is reference Top-of-Atmosphere reflectance data, and B and S B are a spectral range and a spectral response function of the to-be-determined hyperspectral imager respectively;
at step 5.2, based on geometrical conditions of earth observation, the equivalent radiance DN λ of each spectral band at the entrance pupil of the to-be-determined hyperspectral imager is obtained, and a radiometric reference transmission link equation comprises:
DN
_
λ
=
〈
ρ
TOA
(
λ
)
〉
×
E
s
u
n
(
λ
)
×
cos
(
θ
SAT
.
)
π
×
D
S
E
2
wherein DN λ is an equivalent radiance of each spectral band at the entrance pupil of the to-be-determined hyperspectral imager, which is in the unit of w/(m 2 sr μm); ρ TOA (λ) is a mean Top-of-Atmosphere reflectance corresponding to the central wavelength λ of the to-be-determined hyperspectral imager, which has no unit dimension; D SE is an earth-sun distance represented by astronomical unit (AU); E sun (λ) is a solar spectral irradiance of a position corresponding to the central wavelength λ of the to-be-determined hyperspectral imager at the time of satellite imaging; and θ SAT. is a solar zenith angle at the time of satellite imaging.
7. The curve-based radiometric calibration method of the spaceborne hyperspectral imager according to claim 1 , wherein with the constraints of spatial, spectral and temporal resolutions, completing system-level preprocessing of the to-be-determined hyperspectral imager and measuring the radiometric calibration coefficient reference value of the to-be-determined hyperspectral imager on orbit comprise the following steps:
at step 6.1, the mean gray value L λ of each band of representative pixel points is calculated:
〈
L
λ
〉
=
〈
L
λ
〉
total
MN
wherein M and N are a row and a column of representative pixels respectively, with the total number being MN; L λ total is a total gray value of each band λ obtained by traversing representative picture elements of each band, and L λ is a mean gray value of each band of representative pixel points;
at step 6.2, based on the constraints of spatial, spectral and temporal resolutions, the radiometric calibration coefficient reference value is obtained by least square fitting:
Lj = G λ × L λ
wherein L λ is a weighted mean gray value corresponding to the band central wavelength λ, L λ is a mean gray value of each band of representative pixel points corresponding to the band central wavelength λ, and G λ is a radiometric calibration coefficient reference value.
8. The curve-based radiometric calibration method of the spaceborne hyperspectral imager according to claim 1 , wherein based on the radiometric calibration curve accuracy evaluation result, performing fitting on the data points of the radiometric calibration coefficient reference value by fitting and establishing the postlaunch radiometric calibration coefficient curve equation comprise the following steps:
at step 7.1, based on the radiometric calibration curve accuracy evaluation result, a cubic polynomial function is used as the postlaunch radiometric calibration coefficient curve equation:
G λ =p 1 λ j 3 +p 2 λ j 2 +p 3 λ j +p 4
wherein G λ is a radiometric calibration coefficient reference value, j is a detector array row number, and p 1 ,p 2 ,p 3 ,p 4 are coefficients of the degrees of the corresponding items of the polynomial;
at step 7.2, the coefficient of each item of the postlaunch radiometric calibration coefficient curve equation is calculated by the least square method, and the postlaunch radiometric calibration coefficient curve equation is solved;
at step 7.3, the root-mean-square error
RMSE
=
1
n
∑
j
=
1
n
(
M
λ
-
F
λ
)
2
is used as curve accuracy evaluation formula, wherein F λ is a fitting value of the band central wavelength λ, M λ is a measurement reference value of the band central wavelength λ, n is a total wavelength number and j is a detector array row number.
9. The curve-based radiometric calibration method of the spaceborne hyperspectral imager according to claim 1 , wherein based on the postlaunch radiometric calibration coefficient curve equation, performing periodic calculation on the radiometric calibration coefficient reference value of the to-be-determined hyperspectral imager on orbit to obtain an updated on-orbit attenuated radiometric calibration curve comprises the following step:
based on the linear relational expression λ c (j)=g×j+w 0 between CMOS detector row number and central wavelength and the postlaunch radiometric calibration coefficient curve equation, periodic calculation is performed on the radiometric calibration coefficient reference value of the to-be-determined hyperspectral imager on orbit to obtain an updated on-orbit attenuated radiometric calibration curve.
10. A curve-based radiometric calibration system of a spaceborne hyperspectral imager, comprising:
a matching module, configured to, before a to-be-determined hyperspectral imager is launched, match a linear variable filter with a Complementary Metal Oxide Semiconductor (CMOS) detector array to acquire a matching result of a CMOS detector row number;
a correcting module, configured to, in a laboratory dark background environment, perform no-reference radiation source imaging and calibration light source imaging of the to-be-determined hyperspectral imager, and complete additive noise correction and non-uniformity correction of the CMOS detector;
a measuring module, configured to, based on the calibration source light imaging and measured spectral radiance distribution, measure a radiometric calibration coefficient reference value of several conventional spectral channels;
a first establishing module, configured to, based on the radiometric calibration coefficient reference value, establish a prelaunch radiometric calibration coefficient curve equation and determine a radiometric calibration curve accuracy evaluation result;
a second establishing module, configured to, after the to-be-determined hyperspectral imager is launched, establish a radiometric reference transmission link;
a constraining module, configured to, with constraints of spatial, spectral and temporal resolutions, complete system-level preprocessing of the to-be-determined hyperspectral imager and measure the radiometric calibration coefficient reference value of the to-be-determined hyperspectral imager on orbit;
a third establishing module, configured to, based on the radiometric calibration curve accuracy evaluation result, perform fitting on data points of the radiometric calibration coefficient reference value by fitting and establish a postlaunch radiometric calibration coefficient curve equation;
an updating module, configured to, based on the postlaunch radiometric calibration coefficient curve equation, perform periodic calculation on the radiometric calibration coefficient reference value of the to-be-determined hyperspectral imager on orbit to obtain an updated on-orbit attenuated radiometric calibration curve, and obtain an imaging band radiometric calibration coefficient based on the updated on-orbit attenuated radiometric calibration curve.Cited by (0)
No later patents cite this yet.
References (0)
No backward citations on record.